Real-Time Physics Simulation Forum

Hi..

Last week I read several articles about Bounding Volume. It?s really interesting. By the way, is that really necessary to implement BV in hierarchical form or just enough using single BV for each object? Let?s say for virtual environment application in which the application has 100 spaceships undergoing rigid body motion and uses k-Dops as our BV. My question is, do I need to decompose each of the spaceships into several portions hierarchically and each portion will have their own k-Dops? If I don?t want to decompose the spaceship into several portion and just using single k-Dops, is that sufficient to validate collision detection in virtual environment application.

I?m quite confusing since sometimes certain researchers use single BV to validate collision and sometimes they use hierarchical form of BV to certify collision in their application.

Any comments and suggestions are really appreciated.

Best regards,

-ab-
Korea.

This paper by Kenny has some good overview:

http://www.diku.dk/publikationer/teknis ... /02-04.pdf

Thanks Erwin,

I've read that paper and its true the given paper gave me some broad idea about BVH. To me, i'm still curious why BV and hierarchy must be formed as one entity. Is there any research paper totally focus on discussing only BV. To my knowledge, i found at least 3 PhD thesis talking about hierarchy of BV.

From Erleben paper i found that BVH basically covers tree generation, traversal, splitting rules, intersection checking and so on so forth. If i'm not mistaken, there is no discussion about why this shape of BV let's say k-Dops was better then SSV, why sphere shape could not produce optimal fitting compared to parallelepiped shape. May be my question is quite funny. Anyway, let me find the solution from Christer book. So..have a nice day. Thanks very much.